A Rolling Wavelet-Denoised TENET Framework for Analyzing Tail-Risk Spillovers in Crypto-Equity Complex Systems

Abstract

This paper develops a rolling wavelet-denoised Tail-Event driven NETwork (RWD–TENET) framework to examine time-varying tail-risk spillovers between cryptocurrencies and global stock indices. Within each rolling historical window, the framework filters high-frequency disturbances from asset returns and then applies TENET to construct directional tail-risk networks, thereby avoiding look-ahead bias. Using daily data for eight representative cryptocurrencies and seventeen global stock indices from 10 October 2018 to 24 November 2025, we compare networks estimated from raw and denoised returns. Controlled simulations and empirical results indicate that RWD reduces noise-induced distortions while largely preserving extreme-event information and the main temporal structure of asset returns. Consequently, the denoised networks are sparser but exhibit greater edge concentration, stronger average transmission, and higher global efficiency. Dynamic connectedness varies substantially over time and rises sharply during major episodes of financial stress, revealing pronounced cross-market tail-risk contagion. At the group level, global stock indices generally act as risk receivers, whereas cryptocurrencies display stronger risk-emitting behavior, particularly during stress episodes. At the asset level, both risk reception and emission are concentrated in a small number of systemically important nodes, although their roles vary across market states. These findings suggest that RWD–TENET provides a robust and interpretable framework for identifying economically meaningful tail-risk transmission channels and systemic roles in interconnected crypto-equity systems.